<meta http-equiv="Content-Type" content="text/html; charset=UTF-8" />
At the user conference in June at Schott in Mainz, I presented the new Recode function and the option to view texts that differ by less than x percent as the same and to summarize them. The question arose as to whether the x percent refers to the shorter or the longer text. I did a little experiment with texts of different lengths. The shorter text may have as many fewer characters as x percent of the longer text. In my test, I want to choose x with 25, i.e. consider changes that include up to 25% of the characters to be the same. This is how I created the column below. The first text contains 12 characters, 25% of which are 3. Therefore I entered a text with 9 and one with 15 characters
As you can see from the result, the twelve- and fifteen-digit texts are combined, so the maximum permitted difference is determined using the shorter text. This also applies to comparisons involving several similar documents of different lengths.
If you add a 16-digit sequence of numbers, the same grouping is formed as before. 
As you can see on closer inspection, the texts are sorted before the comparison. The first texts that meet the criterion are then grouped, not necessarily those with the least distinction!
The support that this function offers is enormous, and it cannot be used uncritically.
